Stargazer() and starprep() with different robust clustered standard errors [R]

Question

Introduction

I need to replicate this regression table in R: regression table from study

I successfully replicated each individual regression model (1 - 4c) and can display them individually with the stargazer package.

• OLS regression with lm() such as

reg1_2 <- lm(contribution ~ period, data = panel_data_2)

• every model needs to have robust clustered standard errors. Cluster variable: idsess. However, I am using different data sets for every model. Model 1-3 share the same, however the data sets for 4a-4c needs to be modified to exclude some participants or choose different periods. For example:

panel_data_4a <- panel_data_2 %>% 
  filter(idtyp  !=4)

reg4a_2 <- lm(contribution ~ predictedcontribution + belief, data = panel_data_4a)

Problem

Usually, I would use the stargazer command to display the regression table for model (1):

stargazer(reg1,
          se = starprep(reg1, clusters = panel_data_2$idsess, se_type = "stata"),
          type = "html")

Can the command be used so that starprep can use different data sets for computing the robust standard errors? Maybe something like that (I know, this code does not work. This is to make my question a bit clearer):

stargazer(
  reg1_2, reg2_2, reg3_2, reg4a_2, reg4b_2, reg4c_2,
  se = list(starprep(reg1_2, clusters = panel_data_2$idsess), 
            starprep(reg2_2, clusters = panel_data_2$idsess), 
            starprep(reg3_2, clusters = panel_data_2$idsess),
            starprep(reg4a_2, clusters = panel_data_4a$idsess),
            starprep(reg4b_2, clusters = panel_data_4b$idsess),
            starprep(reg4c_2, clusters = panel_data_4c$idsess)),
  type = "html",
  keep.stat = c("n", "rsq")
)

I already tried to...

# Calculate the standard errors for each model
se_reg1_2 <- starprep(reg1_2, clusters = panel_data_2$idsess, se_type = "stata")
se_reg2_2 <- starprep(reg2_2, clusters = panel_data_2$idsess, se_type = "stata")
se_reg3_2 <- starprep(reg3_2, clusters = panel_data_2$idsess, se_type = "stata")
se_reg4a_2 <- starprep(reg4a_2, clusters = panel_data_4a$idsess, se_type = "stata")
se_reg4b_2 <- starprep(reg4b_2, clusters = panel_data_4b$idsess, se_type = "stata")
se_reg4c_2 <- starprep(reg4c_2, clusters = panel_data_4c$idsess, se_type = "stata")

# Extract the numeric standard errors
se_values_reg1_2 <- se_reg1_2$getCov()[upper.tri(se_reg1_2$getCov())]
se_values_reg2_2 <- se_reg2_2$getCov()[upper.tri(se_reg2_2$getCov())]
se_values_reg3_2 <- se_reg3_2$getCov()[upper.tri(se_reg3_2$getCov())]
se_values_reg4a_2 <- se_reg4a_2$getCov()[upper.tri(se_reg4a_2$getCov())]
se_values_reg4b_2 <- se_reg4b_2$getCov()[upper.tri(se_reg4b_2$getCov())]

# Extract the standard errors for reg4c_2 using getCov and extract diagonal
se_values_reg4c_2 <- se_reg4c_2$getCov()[upper.tri(se_reg4c_2$getCov())]
se_values_reg4c_2 <- diag(se_values_reg4c_2)

# Create a list of standard errors
se_list <- list(se_values_reg1_2,
                se_values_reg2_2,
                se_values_reg3_2,
                se_values_reg4a_2,
                se_values_reg4b_2,
                se_values_reg4c_2
                )

# Create the HTML table
stargazer(list(reg1_2, reg2_2, reg3_2, reg4a_2, reg4b_2, reg4c_2), se = se_list,
          type = "html",
          style = "aer",
          title = "Table 2 - Explaining Contributions",
          dep.var.labels = "Dependent variable: Contribution",
          covariate.labels = c("Period", "Predicted contribution", "Belief"),
          column.labels~c("(1)", "(2)","(3)", "(4a)", "(4b)", "(4c)"),
          column.separate = c(1,1,1,1,1,1),
          notes = "<em>Robust standard errors in parentheses.<br>Models 4a to 4c exclude (confused) subjects who, on the basis of the P-experiment,<br>could not be classified according to the FGF classification as either a<br>“free rider,” “conditional cooperator,” or a “triangle contributor.”</em>",
          notes.align = "l",
          keep.stat = c("n", "rsq"),
          model.names = FALSE,
          model.numbers = FALSE
          )